Content Pruning: Deleted 33% of Pages, Clicks Up 37% and Hard Lessons Learned

On a healthcare company's website, there was a Christmas card from 2021 still sending New Year wishes to Google. Google, honestly, never replied. That card was not alone. It had roommates: seven "in case you missed it" posts for people who genuinely missed it, an early registration reminder for a conference that ended two summers ago, and an invitation to a maritime-themed reception—which I can only guess was a wonderful evening for everyone except the XML sitemap.

All of them were indexed. All of them were standing at Google's doorstep with a resume. And that is index bloat—a condition most corporate websites suffer from without realizing it. Nobody sits in a meeting and decides to publish 50 thin event pages. A template did that. Mercury is not in retrograde. The content management system (CMS) is the culprit.

In December 2024, the SEO director in this story did what every SEO professional threatens to do but rarely gets approved: she crawled 600 URLs and marked 198 pages—one-third—for a "one-way ticket". Those pages had earned a total of 144 clicks. That is 3.1% of the site's total clicks, roughly what the homepage does before lunch. And here is what happened next. All of it. Even the embarrassing parts.

The Shocking Result: Delete One-Third of the Website, Clicks Increase by 37%

Let's start with the numbers that made many sit up straight. In the four months following content pruning, Google Search Console data showed:

Metric	December 2024	April 2025	Change
Search Console Clicks	5,710	7,840	+37%
Search Console Impressions	247,000	368,000	+49%
GA4 Organic Sessions	10,404	16,103	+55%
Ranking Keywords (Semrush)	4,348	6,742	+55%
Top 10 Keywords (DataForSEO)	264	271	+3%
GA4 Paid Search Sessions	11,863	182	-98%

Looking at this table, the first three rows look as pretty as a dating profile. But the final row—paid search dropping by 98%—is why any pruning case study with only an upward-trending chart deserves skepticism. Even the chart the author herself might draw from this data.

The most important point the author affirms: nothing broke. She stated that this is the single result in the article she is willing to defend in court. Deleting one-third of the crawled URLs did not cause any measurable loss in rankings. That is the foundation for every conclusion that follows.

198 Deleted URLs: What Was Actually Cut?

This was not a blind massacre. Four types of pages were removed, and not a single one of them was written for a human with a real question.

Page Type	URL Count	Action
Old event promos (speaker previews, early bird reminders, boat parties)	50	301 redirect to event page, then noindex
Holiday cards, ICYMI roundups, old news, 2020 pandemic posts	43	301 redirect to resource page, then noindex
Client logo pages, each 94–156 words long	51	Unpublished and redirected
Thin podcast stubs, syndicated posts, near duplicates	24	Noindex, nofollow

Among the 198 cut pages, 140 had never earned a single click. Not during slow months. Never. A second wave executed a month later removed another 60 orphaned URLs, mostly taxonomy archives that the CMS had quietly propagated in the basement.

And here is a sobering detail about those event pages: when asked why there were so many event pages, the web manager replied that for an event to appear on the site, the CMS forced the publication of a full page for it. "Complex and silly," she wrote. That is the exact nature of index bloat. Nobody decided to do that. A template did.

How to Decide to Prune: Read Every URL, Use No Traffic Thresholds

The methodology here was nothing flashy: read every URL. Each page had a row in a spreadsheet: status code, indexability, word count, internal links, near duplicate score, alongside clicks, impressions, and positions from Search Console. Then each row was assigned a description and placed into one of four tabs: Redirect, Noindex, Keep, Questions.

What is remarkable is that there was no traffic threshold in the process. The number that broke all traffic thresholds: out of 386 retained pages, 213 also had zero clicks. If the rule was "no clicks, no mercy", more than half of the surviving pages would have been wrongfully deleted. Those were product pages, case studies, and privacy policies—things nobody searches for, but the legal department wants to keep.

Pruning is like the art of bonsai: cut small, and look at the whole tree first. Clicks only tell you where to look. They do not tell you what to do. Four questions make the decisions:

• Was this page ever created for search? A holiday card was not. Gone.
• Thin now, but valuable later? Noindex and put on the rewrite list.
• Does the business need it to exist? If yes, it lives. Read the next section before arguing.
• Is there a closer page? If two pages cover the same topic, choose one and point the other to it.

And a working rule that matters more than all four questions combined: Redirect first, then noindex or unpublish. Two steps. In that order. Skipping this order was the biggest mistake in the entire project.

Hard Lessons Learned: 51 404 Pages and Indispensable Takeaways

This is the part that makes this case study more believable than any report filled only with pretty metrics. The author lists six mistakes, in the order that still haunts her.

1. Failing to Take "After" Screenshots

She had indexed page counts from three weeks prior to pruning and from 22 months afterward. Nothing in between. Search Console only retains 90 days of index history and 16 months of performance data. The lesson: export reports on day one, and repeat on days 30, 60, and 90. Google will not keep memories for you.

2. Unpublishing Before Redirects: 51 404 Pages

The client logo pages were unpublished before redirects were created. For a time, all 51 pages returned 404s. The author's notes across each row of data read like a ransom note: "confirmed unpublished, currently 404, please redirect." Two steps. Right order.

3. Mistakenly Deleting a Page Scheduled for Content Roadmap

An executive Q&A was marked for redirect and noindex while it was scheduled for a refresh. The spreadsheet note simply read: "So sorry! This was a mistake." The page was restored to the index within days, only because a human happened to notice.

4. Redirecting to a Hub Page Instead of the Best Match

50 event posts were redirected to the general events page. 43 went to the resource page. 51 client pages went to a generic "why choose us" page. The final destination page earned 85,200 impressions and 50 clicks over 16 months at an average position of 7.3. At first glance, the author blamed the redirect. But when pulling the queries, the biggest query—29,596 impressions and exactly one click—was the company's bare abbreviation. The redirect was not at fault. It was just lazy. Nobody searching for the story of a specific hospital wants to land on a generic "why choose us" page.

5. Redirects Are Debt

By September 2026, 278 out of 923 crawled URLs were 301 redirects, and 3,390 internal links still pointed to them. They removed the pages from the index but left links scattered throughout the house, like an ex-boyfriend's hoodies.

6. Templates Keep Publishing

Months after deleting 50 event posts, a brand new event post sneaked into the site's top 5 landing pages. The next conference produced another batch. They mopped the floor but left the faucet running.

The Naked Truth: 55% Organic Growth Wasn't Entirely Due to Pruning

This is the part most SEO case studies omit. The author did not.

During the same quarter as the pruning, paid search was turned off. GA4 recorded 11,863 paid search sessions in December 2024 and only 182 in April 2025. On a website where most users search for the brand name, those visitors did not disappear. They switched to clicking organic results.

Let's do the painful math in the article: organic increased by 5,699 sessions. Paid search dropped by 11,861 sessions. Combine the two channels, and total search sessions decreased by 27%, from 22,267 down to 16,285. So that beautiful 55% organic growth? A large portion of it was ad budget wearing a different outfit. That is the chart nobody puts into a case study. And the author included it.

What the author could claim is smaller but more useful: she deleted one-third of the URLs and lost no measurable rankings. Semrush's site health score jumped from 69% to 79% the month changes were deployed, then gave back nine points in February. But top 10 keywords did not surge immediately. They stood at 264 in December 2024 and 271 in April 2025. Only after the rewrites had been live for several months did the number climb to 338 in December 2025—a 28% increase over pre-pruning baselines.

Pruning cleaned out the room. New content did the actual work.

Podcasts: From 90 Words to Position 9.0, But Clicks Dropped

The most interesting part of the entire project lay in the 24 podcast pages. On paper, they were the worst content on the site: a title, an audio player, a two-sentence intro. The word count was enough to embarrass a fortune cookie. But they were also the best content the company owned.

Each episode was about 30 minutes of interviews with a healthcare system leader, a chief experience officer, or a doctor—people explaining precisely how they solved a problem that audiences were searching for on Google. Real names. Real hospitals. Real numbers. The exact kind of direct expertise every SEO presentation begs clients to produce, and here it was produced, locked inside an MP3 file that no crawler and no answer engine could hear.

Deleting those pages would have been like donating a coat to charity while leaving $10,000 in the pocket. So the author kept the pages and built out the content above them. Each episode became an in-depth, long-form article at the old address: expert arguments, action steps, quotes, and data, all written out, with the audio remaining for those who prefer listening at 1.5x speed.

Year-over-year comparison results: the podcast section's average Search Console position went from 17.3 down to 9.0. Podcast pages with impressions went from 22 to 24. But clicks dropped from 195 to 112. Down. The author did not hide it. She explained: this is search in 2026, where 220 of this site's keywords now trigger AI Overviews (compared to 16 a year prior), and a top 10 ranking is sometimes just a consolation prize. Part of the position gain also stemmed from Google's reporting shift in September 2025, which stopped counting many deep rankings and flattered everyone's averages.

Even so, the author maintained she would make the exact same decision tomorrow. Because the alternative was 24 pages of nothing, and you cannot get cited by an answer engine thanks to a blurb and a play button.

Indexed Pages Down 57%, Top 10 Up 65%: The Long-Term Picture

If you could only choose one image for the entire project, this is it. Search Console showed 1,412 indexed pages in September 2024 and 1,297 pages on December 2, 2024, right before pruning. By October 2026, the count was down to 559 pages—a 57% decrease. Another 1,369 URLs sat under the "Excluded by noindex tag" section, all there intentionally.

The number 559 did not come from pruning alone. It also included a subsequent rule that noindexed 374 old PDF files, a hosting migration, and nearly two years of phases two through four. But it demonstrates the trajectory: the website had less than half as many indexed pages as before, and 65% more keywords in the top 10 than before cuts began: 435 in October 2026, up from 264.

That was not a straight upward line. It was a two-year process involving rewrites, waiting periods, and corrected mistakes. But it proved one thing: a smaller, leaner website can outrank a bloated website packed with pages nobody needs.

Content Pruning Checklist for Vietnamese SEO Practitioners

These lessons were learned with real money and time. Here is what to do if you are considering content pruning for your website.

1. Read Every URL, Don't Just Look at Aggregate Metrics

Do not use a single click threshold to decide the fate of every page. In this case study, 213 out of 386 retained pages also had zero clicks. They survived because the business needed them, legal needed them, or they would hold value after being rewritten.

2. Redirect First, Noindex or Unpublish Second

This is the rule that caused 51 404 pages. Two steps. Right order. Do not skip either.

3. Take "Before" and "After" Screenshots on Day One

Export index reports, crawl stats, and clicks for all redirected landing pages right at the start. Repeat on days 30, 60, and 90. Search Console does not keep your data long enough for you to regret it.

4. Fix Internal Links in the Same Batch

A redirect without updated internal links is just a messy forwarding address. If 3,390 links still point to redirected URLs, you are leaving behind technical debt.

5. Check Content Roadmaps Before Pruning

Zero clicks paired with a refresh schedule next month does not make a pruning candidate. That is a page waiting to be revived.

6. Ask What Lies Inside Thin Pages

A 90-word page wrapped around a 30-minute expert interview is not thin. It is an unopened treasure chest. Do not delete it. Open it up.

7. Write Rules to Stop Bloat

Pruning without fixing the template is like mopping the floor while leaving the faucet running. The author's rule, ultimately: one event page, updated annually. If dedicated promo pages are needed, they are noindexed from birth.

Pruning Doesn't Create Growth; It Clears Space for Growth

Content pruning is not a miracle cure. It does not turn a weak website into a strong one overnight. In this story, it removed one-third of the pages at the cost of 3% of clicks, caused no measurable ranking losses, and sparked a conversation more important than any redirect: why does this website keep producing pages nobody wants, and what are we hiding inside pages we thought were worthless?

What I learned after 20 years in SEO is not how to get more pages. It is how to make every single page deserve to exist. A smaller, better-cared-for website will always beat a bloated website featuring thousands of unread, uncited, unneeded pages.

And there is one final detail in this story that makes me smile: last week, the author pulled the site's "not found" report—58 URLs. Seven needed fixing, and 41 had never been pages at all. And one podcast episode was stuck in a redirect loop, two URLs pointing to each other indefinitely, each waiting for the other to go first. Someone will have to break it. Probably her. Probably on a Tuesday.

SEO work never truly ends. It just shifts from major overhauls to minor daily edits.

How many pages on your website do you know for sure nobody needs, yet they remain because "they were already created" or out of "fear of deleting something important"? If tomorrow you faced the decision to delete one-third of your URLs, what holds you back—fear of losing rankings, or the reality that you have never truly looked at each page to see if it deserves to exist? Share your story below, because in the war against index bloat, those who dare face the truth about their website are the first to find a way out!